add 32 export in raw

[kerautret]

just to export 32 bits images

[dcoeurjo]

So you can consider DGtal::uint32_t as specialization type instead if unsigned int

[kerautret]

yes ok

[dcoeurjo]

Without the raw32 reader, it's a bit hard to check possible issues with the endianess.

[dcoeurjo]

any update ?

[kerautret]

I look it

[kerautret]

I am working on Raw32 integration in generic reader not yet finished

[kerautret]

@dcoeurjo should be good now : we have both reader and writer 32 bits for raw ;à

[dcoeurjo]

maybe it's dangerous here... We using genericwriter on value type T, the behavior would be:

• export as raw8 with cast function from T -> char
• except if T==uint32_t which would be exported as raw32

This is maybe a bit complex/strange for the user... I would say: in the generic wirter/reader, we only consider raw8 export. If the user wants raw32, he/she has to explicitly use the Raw32 method of the raw writer.. what do you think ?

[kerautret]

I am not sure that complex for user:

• If the user has uint32 image, by default the generic writer will write it with 32 bits support.
• But if strangely enough, the user prefers to export in raw8 he can specifically call the raw8 of the RawWriter
• if the user has a 8 bits image, the raw8 will be called an no pb.

But perhaps you are right just if a user use a 32 bits image and just want export it in pgm it will compile but wll finished with an error message in the execution. Perhaps we can just indicate to the user to change the image mode to export in pgm ?

[dcoeurjo]

okay but what is the behavior for 'unsigned int' ? raw32 or raw8 ?
I don't know if the compiler is such that unsigned int == uint32_t

But that's weird case I admit..

Have you make it clear in the doc ?

[kerautret]

perhaps it is dependent of the system, for me unsigned int = uint32 but since it only specialized with uint32 it should be fine.
I add some details in the doc

[kerautret]

@dcoeurjo done

[dcoeurjo]

ok merging